The number of new heroin users and problems associated with heroin use have increased steadily over the past several years. Methadone maintenance is currently the most effective treatment for opioid dependence but it has limitations and is controversial. An alternate pharmacological strategy, naltrexone maintenance, has currently limited usefulness due to poor compliance and low patient acceptability. Improved treatment approaches, including novel medication as well as pharmacological and behavioral augmentation strategies for naltrexone are greatly needed. Preclinical studies and preliminary clinical observations support the use of NMDA receptor antagonists in the treatment of opioid dependence. These compounds inhibit drug-conditioned responses that play a role in relapse, reduce opiate self-administration, and attenuate opiate withdrawal in laboratory animals. In humans, NMDA receptor antagonists reduce signs and symptoms associated with opiate withdrawal, and reduce subjective effects of heroin and heroin craving. The goal of this five-year study is to test the efficacy of memantine (a noncompetitive NMDA receptor antagonist) as an adjunct to the maintenance treatment with naltrexone in detoxified heroin-dependent individuals. In the proposed trial, one hundred and sixty five heroin-dependent patients who completed detoxification will be randomly assigned to one of three conditions (1) Naltrexone + Placebo;(2) Naltrexone + Memantine 30 mg bid, and (3) Naltrexone + Memantine 60 mg bid (N=55 per group). Naltrexone will be dispensed three times per week in the clinic, while memantine or placebo will be taken at home. In addition, patients will receive twice weekly psychosocial intervention that will include motivational interviewing and cognitive-behavioral relapse prevention. The goal of psychosocial intervention is to improve compliance with medication and maintain abstinence. A double-blind trial will last twelve weeks with assessments at baseline and at each appointment three times per week. Repeated assessments will also be completed one, two, and three months following the end of treatment. Primary outcome measures will be retention in treatment by the end of the study and heroin abstinence in the final four weeks prior to study endpoint. The primary aim is to test the efficacy of memantine in reducing early attrition and improving outcome in opioid-dependent individuals maintained on naltrexone.